Cocaine Addiction: Psychology and
Neurophysiology
FRANK H. GAWIN
Cocaine was considered incapable of producing depen-
dence in 1980 but was recently proclaimed the drug of
greatest national health concern. Recent clinical and pre-
clinical investigations demonstrate that cocaine produces
unique abuse and withdrawal patterns that differ from
those of other major abused drugs and suggest that
long-term cocaine abuse produces neurophysiological al-
terations in specific systems in the central nervous system
that regulate the capacity to experience pleasure. It will be
necessary to develop clinically pertinent research models
before these findings can be considered definitive, but
these evolving ideas have already led to applications of
promising experimental treatments for cocaine abuse.
A DDICTION TO COCAINE DOES NOT CAUSE GROSS PHYSIO-
logical withdrawal symptoms. However, the addictive con-
sequences of cocaine use have forced reconsideration of the
importance attributed to "classic" drug abuse constructs, such as
physiological withdrawal, tolerance, and physical dependence, in
drug and alcohol addiction. These constructs are based on easily
quantitated gross physiological alterations in parameters such as
heart rate or blood pressure and on easily observed symptoms of
physical illness. Their central position in drug abuse theory and
research, although questioned, has been unchanged for decades;
drug dependence has thus largely been considered an avoidance of
physical discomfort on withdrawal.
Pharmacological agents have been developed that reverse the
gross physiological withdrawal symptoms produced by benzodiaz-
epines, alcohol, and opiates. However, these agents, fully assessed in
recent clinical treatment trials, have not solved the problem of
addiction to these drugs. The facts that cocaine produces no gross
physiological withdrawal symptoms and that, in withdrawal from
other abused drugs, agents reversing such symptoms show limited
efficacy, demonstrate that subjective experiences or symptoms other
than physiological discomfort are crucial in addiction to cocaine and
to other substances of abuse.
Clinical and preclinical investigators are now exploring how
psychological symptoms in drug withdrawal, particularly unpleasant
mood states and cravings for drug euphoria, maintain chronic drug
addiction. Such symptoms may be mediated by neurophysiological
adaptations that are attributable to repeated perturbations of the
central nervous system (CNS) by abused drugs. In cases where these
adaptations occur in brain systems that regulate only processes
lacking physical expression, as appears to be the case with cocaine,
addiction may have been erroneously considered as purely "psycho-
logical" rather than "physiological" and had erroneously been
The author is in the Department of Psychiatry, Stimulant Abuse Treatment
Research, University of California, Los Angeles, CA 90024.
1580
presumed to pose less severe treatment problems. The perception
that cocaine produced purely a psychological problem wrongly
implied that it was appropriate to limit treatment to psychological
therapies and suggested there was no role for neurophysiological
interventions with pharmacological agents. These drawbacks of
prior "classic" tenets in the drug abuse field has recently led the
World Health Organization to discard these terms in favor of the
more physiologically precise term "neuroadaptation" (1) and has led
the American Psychiatric Association to define drug dependence by
Downloaded from www.sciencemag.org on October 5, 2012
using behavioral criteria, rather than only physiological criteria (2).
The epidemic ofcocaine abuse may therefore provide one positive
legacy-forced recognition ofthe complexity ofsubstance abuse and
new conceptualizations of and research on addiction. In this article,
I review the recent concordance of clinical and preclinical findings
about long-term cocaine exposure that suggest cocaine causes a
neurophysiological addiction. Corresponding advances in the treat-
ment of cocaine abuse are noted, but these are discussed fully
elsewhere (3, 4). Further, this article will describe the methodolog-
ical limitations of current preclinical and clinical research to illustrate
where conceptual or technical methodological advances are now a
prerequisite to systematic advancement of research, theory, and
treatment of the addictions.
The Clinical Characteristics of Cocaine
Addiction
We are in the largest cocaine abuse epidemic in history. In the
United States alone, one to three million cocaine abusers are
estimated to be in need of treatment (5), or up to six times the
number of heroin addicts. In the 1890s, cocaine use also surged and
was considered safe by clinicians as illustrious as Sigmund Freud (6).
Use declined after its dangers became well known (7). This recurred
in the 1920s. In the early 1950s and late 1960s, abuse of the similar
stimulants, amphetamine and methamphetamine, resulted in the
same pattern (8). In 1980, cocaine was again considered a safe,
nonaddicting euphoriant (9). Historical descriptions of cocaine
dependence were dismissed as exaggerations similar to marijuana
reports from earlier eras. Clinical research on cocaine abuse that used
systematic and objective techniques did not exist, and this led to the
miscomprehension that cocaine dependence did not exist. Cocaine
use then exploded. Almost one of two Americans between 25 and
30 years of age has tried cocaine. A powerful new route of
administration has appeared; cocaine smoked as the free base, also
called "crack," is as addictive as intravenous injection, without the
stigma or infectious dangers of injection (10).
Individuals who are in treatment for cocaine abuse report that 2
to 4 years intervene between the initial exposure to cocaine and the
development of addiction (11); this interval has delayed reports of
and cocaine's adverse effects at the beginning of epidemics and, com-
bined with reports of apparently controlled initial use, promotes
SCIENCE, VOL. 251
illusions of safety. Early in the course of cocaine epidemics, the
public, scientists, respected texts, and astute clinicians have all fallen
victim to this illusion, thus enabling the repeated emergence of
cocaine abuse.
Psychological effects of cocaine: initial use. In both reports from
individuals addicted to cocaine and human laboratory investigations
(in which cocaine is administered to nonaddicted cocaine users),
cocaine initially induces profound subjective well-being together
with alertness (6, 7, 12, 13). The fundamental effect of cocaine is the
magnification of the intensity of almost all normal pleasures. The
environment takes on intensified but nondistorted qualities. Emo-
tions and sexual feelings are enhanced (6, 13). Self-confidence and
self-perceptions of mastery increase, so anxiety is initially decreased.
Social inhibitions are reduced, and interpersonal communication is
facilitated. Satiation of appetite occurs, so pleasures associated with
eating are not enhanced (11, 12).
Human laboratory studies of cocaine's effects are more reliable
than retrospective self-reports by cocaine users, but they are serious-
ly limited as aids in understanding cocaine abuse because there are
multiple disparities between the methods used in such experiments
and the characteristics of actual street cocaine use. Maximum
individual laboratory doses are 10% of the maximum doses typically
reported by street abusers (12). The maximum duration of admin-
istration in human experiments is 5 hours (14), compared to street
binges lasting as long as 200 hours. Cocaine abusers usually
coadminister other drugs (alcohol, heroin, and marijuana) to de-
crease unpleasant components of the cocaine experience, such as
anxiety, that admix with euphoria. The effects ofcoadministration of
these drugs, of their relative dosages, and of varying the time
patterns of their administration have not been studied. Further,
because administration of cocaine to a cocaine abuser in need of
treatment has been considered unethical, the subjects in such studies
are not in need of treatment for cocaine abuse, have less severe
histories of cocaine abuse, and may often instead have substantial
histories of abuse of other substances. These facts raise the possibil-
ities that any neuroadaptation pertinent to cocaine dependence is
not present, or that neuroadaptation to other abused substances
might be present, or both. Finally, the effects of nonpharmacological
factors that interact with the effects of cocaine have been largely
ignored. Variations in the environmental context of cocaine admin-
istration, such as familiarity with the administration environment,
the spectrum and pleasantness of available activities, degree of
interpersonal isolation, degree of perceived safety, presence or
difficulty of tasks, perceived medical risk, and many others may all
affect cocaine's use and effects but have not been studied. Such
studies are fundamental. For example, it is unknown whether the
distraction that is required for reporting the subjective effects of
cocaine to investigators might substantially alter the experience of
cocaine euphoria.
Psychological effects ofcocaine: addiction and dependence. On the basis
of National Institute of Drug Abuse estimates, only about 10 to
15% of those who initially try cocaine intranasally become abusers.
Some individuals who experiment with cocaine cease use immedi-
ately, typically describing overwhelming anxiety rather than eupho-
ria as cocaine's main effect. Some who do experience intensified
euphoria cease use because of extreme expense or unavailability.
Others, before addiction is fully developed, fear imminent loss of
self-control over cocaine use and are able to cease use.
Individual differences in the progression to addiction could
provide crucial insights for prevention and treatment. However,
reliable predictors of later heavy abuse have not been found in light
cocaine users who have not yet developed addiction; a cocaine
"addictive" personality has not been identified; and no objective,
systematic assessments of the natural longitudinal course of cocaine
29 MARCH 1991
use and abuse exist (3).
As cocaine addiction develops, a transition to high-dose, long-
duration bingeing occurs, in which the intensely pleasurable effects
are experienced alone, and increasingly apparent negative contin-
gencies go unrecognized (3, 6, 10, 15). Users readminister cocaine
every 10 to 30 min. Such compulsive, uncontrolled binge use begins
when availability and dosage escalate (for example, as a result of
increased funds or improved supply sources) or when a switch to
rapid, higher intensity administration routes occurs (from intranasal
use to smoking or intravenous injection) (3, 15-17). During such
binges, there are numerous periods of extreme euphoria, and vivid
memories are formed that are later contrasted with current dyspho-
rias to produce cocaine craving (3, 16). Several days of abstinence
often separate binges; users average one to seven binges per week,
each lasting from 4 to 24 hours (3, 16). Unlike the case for alcohol
or opiate users, the absence of a daily use pattern in a cocaine user
does not indicate decreased impairment, and it may indicate the
opposite (3).
Continuous, rapid, cocaine self-administration also occurs in
animal experiments in which unlimited access to intravenous cocaine
Downloaded from www.sciencemag.org on October 5, 2012
is provided (12). The animals die within 14 days. If access is limited,
an animal will press a lever thousands of times for a single cocaine
dose (12). Human cocaine addicts report that virtually all thoughts
are focused on cocaine during binges; nourishment, sleep, money,
loved ones, responsibility, and survival lose all significance (3, 16).
Supplies of cocaine are drawn on until they are exhausted. Limita-
tions on drug access, including the high price of cocaine and legal
limitations on distribution, regulate human cocaine use and may
prevent human cocaine use from more frequently mimicking animal
free-access experiments in producing death. A low-intensity parallel
to early, controlled use in humans has not been described in animal
self-administration studies, probably because most self-administra-
tion experiments in animals utilize large boluses and intravenous
administration, bypassing an initial phase of low-intensity use and
the binge transition (3).
Cocaine Abstinence Symptoms
A triphasic pattern of symptoms characterizes the response to
abstinence from cocaine (16); this pattern dispels the perception that
cocaine use produces no withdrawal. This abstinence sequence,
initially described on the basis of clinical observations in 30 outpa-
tients, has been verified in one large-scale outpatient study and some
(but not all) assessments in hospitalized cocaine abusers, as well as in
animal models and positron emission tomographic (PET) studies in
humans (18). The symptoms of cocaine abstinence are more com-
plex and subtle than those previously associated with drug with-
drawal, and researchers cannot be certain of the characteristics of
cocaine abstinence until methodological obstacles, delineated below,
can be surmounted. Cocaine abstinence is shown schematically in
Fig. 1.
Phase one-crash. Immediately after cessation of a cocaine binge,
there is a "crash" of mood and energy (3, 10, 16). Cocaine craving,
depression, agitation, and anxiety then rapidly intensify, and, in over
half of abusers seeking treatment, suspiciousness and paranoia are
prominent (14, 19). During the next 1 to 4 hours, cocaine craving
decreases and is supplanted by mounting exhaustion and craving for
sleep; further use is then often strongly rejected, unlike in the
parallel situation after several hours of opiate, sedative, or alcohol
withdrawal (3, 16). Abusers administer marijuana, sedatives, opi-
ates, or alcohol to induce sleep, and after sleep begins, they
experience intense hypersomnolence, with electroencephalographic
changes characteristic of sleep deprivation (20). During brief awak-
ARTICLES 1581
enings, hyperphagia occurs. The hypersomnolence can last several
days, after which mood normalizes.
The crash has been confused with withdrawal (8, 10), but the
crash parallels an alcohol hangover, not the withdrawal associated
with chronic opiate or alcohol administration (3), and although
crash symptoms prolong cocaine binges, they do not maintain
long-term cocaine abuse.
Phase two-withdrawal. In addiction to cocaine, continued use is
induced by a protracted dysphoric syndrome-including decreased
activation, anxiety, lack of motivation, and boredom, with markedly
diminished intensity of normal pleasurable experiences (anhe-
donia)-that emerges shortly (0.5 to 4 days) after the crash. This
hedonically limited existence, when contrasted to memories of
cocaine-induced euphoria, induces severe cocaine cravings, resump-
tion of use, and unceasing cycles of recurrent binges (3, 16). Because
it is directly related to craving and resumption of use, this with-
drawal phase parallels withdrawal from other abused substances,
except that there is an absence of gross physiological changes (16).
These symptoms are far more subtle than those of the crash and
went unrecognized by early observers. They are not constant or
severe enough to meet the criteria for major psychiatric mood
disorders.
The existence of a withdrawal phase is indicated by broad clinical
and preclinical observation and consensus (3, 15, 18). However,
precise and objective quantitation of anhedonia and similar dyspho-
ric symptoms is problematic. Available psychiatric and psychological
rating scales are not sensitive to these subtle mood components. For
example, no scale or technique assessing anhedonia has been vali-
dated in human substance abusers (21). Further, quantitation of
anhedonia requires interactive assessment with a pleasurable probe
because anhedonia is the absence of an appropriate mood response
to a pleasurable event. In addition, there are extensive individual
differences in the perception of what stimuli are pleasurable: there
are few or no events that can serve as plausible human research tools
and that uniformly induce pleasure. In animals, however, elegant
methods involving behavioral or electrophysiological responses to
electrical, pharmacological, or behavioral stimulation have been
developed that allow study of reward responses (reviewed below),
and complementary preclinical data exist to support the hypothesis
that impairment of reward responses occurs after chronic cocaine
abuse.
There are established and validated measurement techniques for
anxiety, unlike for anhedonia. Experimental cocaine administration
Crash
r
Phase Phas 2
YKhdmal
Phase 3
Enco
9 houm
to 4 das to 10wees kcl teflk
Fig. 1. Cocaine abstinence phases. Duration and intensity of symptoms vary therapies targeted at conditioned cocaine craving
on the basis of binge characteristics and diagnosis. Binges range in duration
from under 4 hours to 6 or more days. High cocaine craving in phase 1 plausible and could be screened by assessment of the effect of
usually lasts less than 6 hours and is followed by a period of noncraving with pharmacological agents on cocaine-seeking behavior that has been
similar duration in the next subphase (middle phase 1). Substantial craving
then returns only after a lag of 0.5 to 5 days, during phase 2. [Reprinted linked to
from (16) with permission, Arch. Gen. Psychiatry]
1582
in animal models and anxiety inventory ratings in humans have
confirmed the slightly delayed emergence of anxiety during cocaine
withdrawal (18).
Cocaine abusers desiring to cease use are cognizant, when in
withdrawal, of the adverse consequences of continued cocaine use.
They are usually able to transiently withstand this anhedonic
dysphoria, until they are presented with a conditioned cue. The cue
superimposes a second, more transient dimension of craving
(evoked craving) on anhedonic craving, and, most often, cocaine use
then resumes. Conditioned cocaine craving is described as pulsatile,
lasting only minutes or hours. Cocaine and the amphetamines are
the most potent reinforcing agents known, and as such they produce
intense classical and operant conditioning (12). Such cravings
appear after the appearance of varied, idiosyncratic objects or events
that were temporally paired with prior cocaine intoxications, and
these appearances are experienced as partial memories of cocaine
euphoria. Cues can include mood states (positive as well as nega-
tive), specific persons, locations, events or times of year, mild
alcohol intoxication, interpersonal strife previously soothed by
cocaine euphoria, or abuse objects (for example, money, white
powder, glass pipes, mirrors, syringes, and single-edged razor
Downloaded from www.sciencemag.org on October 5, 2012
blades) (3, 16, 22).
Initiation of a cocaine binge thus depends on an interaction
between drug availability, environmental stimuli (conditioned cues),
and the withdrawal status of the dependent abuser. The time course
of binge resumption fluctuates within and between cocaine-depen-
dent individuals, to an extent much greater than in opiate or alcohol
withdrawal, and depends on the weight of these three factors.
Manipulation of these factors has thus become the focus of initial
medical and societal interventions to control cocaine addiction.
Recent treatment advances focus on decreasing withdrawal dyspho-
ria pharmacologically or on extinguishing conditioned craving by
using graded exposure to cocaine, and public policy advances in law
enforcement could decrease availability of cocaine.
Although the interplay of these factors has been well described in
cocaine-dependent humans, the interactions of these factors have
not been studied in preclinical experiments in animals. Excellent
models exist for each of the factors. Investigators could model
availability of cocaine by changing the parameters of work (change
in pattern or amount of lever-pressing required for a stimulant dose)
or by producing impediments to access (delivery of aversive stimuli).
Conditioned associations, created by pairing of experimental events
with cocaine or amphetamine administrations, have been studied,
isolated from these other factors, quite extensively. Finally, with-
drawal status could be modeled by regulation of the chronicity,
administration route, and pulse pattern of cocaine administration,
and craving can be modeled by assessment of avidity of responding
(point of response cessation after reward ceases) and frequency of
choosing alternative nondrug reinforcers (food, sex, social interac-
tion), after long-term cocaine treatment in animals that precisely
duplicates human use patterns.
Thus, the cocaine disease process could be precisely reconstructed
in animals, particularly regarding administration patterns and aver-
sive stimuli, but animal models designed to mimic human cocaine
administration characteristics have not been used. This is ironic
because the unavailability of precise animal models that impedes
rapid advances in other psychiatric and medical disorders is avoid-
able in studies of cocaine abuse. For example, potential pharmaco-
have not appeared
in clinical or preclinical research. Such treatments are nonetheless
conditioned cues in animals treated with cocaine for long
periods of time in a binge pattern like that seen in humans. Similarly,
SCIENCE, VOL. 251
the relative importance of conditioned craving and withdrawal
craving at varied stages and degrees of abstinence, which is currently
limited to clinical conjecture, could be modeled. Differential and
specific treatment effects are also likely to appear according to the
stages and degrees of abstinence at the time a treatment technique is
applied (for example, isolation from or exposure to cocaine avail-
ability, and pharmacotherapy at the time withdrawal anhedonia is
present). Optimal intervention timing and interactions between
treatments could be precisely modeled and rapidly evaluated. Final-
ly, public policy could be based on experimental findings modeled in
animals, but it is not. For example, the effect of decreasing the
availability of cocaine on increasing work for cocaine (the human
equivalent to this work is often criminal activity) could predict how
the crime rate would change after an increase in cocaine prices. At
present, vast government expenditures occur (on the basis of the
presumption that increased price would decrease use and crime), in
the absence of guidance from such plausible experiments.
If cocaine abusers sustain abstinence, anhedonic symptoms lift
within 2 to 12 weeks (16). Animal studies show that behavioral
depression occurs after stimulant withdrawal for a similar time
period (23). On the basis of clinical observations, symptom severity
and duration depend on the intensity of the preceding months of
chronic abuse and on the presence of predisposing psychiatric
disorders, which amplify withdrawal symptoms (16). Conversely, in
infrequent cocaine users without psychiatric disorders, withdrawal
may not occur. High-intensity binge cocaine use and coinciding
neuroadaptation may be required before withdrawal occurs.
Phase three-extinction. After resolution of withdrawal anhedonia,
intermittent, conditioned cocaine craving can still emerge (16)
months or even years after the last cocaine use. Lasting cocaine
abstinence depends on experiencing intermittent, conditioned crav-
ing without relapse. The previous consistent pairing of cues with
cocaine euphoria then does not occur, and craving is gradually
extinguished (16, 22). Relapse caused by classically conditioned
craving and withdrawal has been described in opiate and nicotine
withdrawal. In cocaine dependence, however, conditioned craving
may be more intense than in other addictive disorders. This is not
surprising because cocaine is such a powerful reinforcer in animal
models and because of the relation between strength of reinforce-
ment and magnitude of classical conditioning.
Short-Term Neurochemical Actions of
Cocaine
Cocaine and the amphetamines are thought to produce pleasure
or reward by increasing neurotransmission in mesolimbic or meso-
cortical dopaminergic tracts, or both, in the brain (24). In animals,
electrical self-stimulation of these pathways produces reward-seek-
ing behavior that mirrors that of cocaine self-administration. Both
electrical self-stimulation of these pathways and cocaine increase
extracellular dopamine concentration in brain nuclei that control
reward behavior. Lesions in these pathways block cocaine effects,
and dopamine receptor blockers attenuate cocaine effects (25).
Whether cocaine also directly influences efferent or afferent connec-
tions is less clear.
Cocaine inhibits reuptake of the neurotransmitters norepineph-
rine and serotonin, as well as of dopamine. It binds to the dopamine
transporter labeled by mazindol and to imipramine binding sites on
serotonergic neurons (26). It is a local anesthetic (26) and also
increases catecholamine receptor sensitivity (27). In contrast to
abused substances like heroin, cocaine does not directly activate
enkephalinergic receptors, but may indirectly influence these sys-
29 MARCH 1991
tems. Cocaine also affects neurotransmission in histamine, acetyl-
choline, and phenethylamine pathways.
None of these neurotransmitter actions are solely responsible for
cocaine euphoria because each is also produced by other pharmaco-
logical agents that do not produce euphoria, are not self-adminis-
tered by animals, and are not abused by humans (3). Thus, although
the rewarding properties of cocaine clearly require activation of
dopaminergic systems, the molecular mechanisms involved and
whether activation is due only to direct cocaine effects on dopam-
inergic neurons or to simultaneous collateral actions on other
neurotransmitters is uncertain (3, 28). Several lines of evidence
support the hypothesis that a simultaneous interaction between
catecholamine and serotonergic systems is fundamental to cocaine
reward and euphoria. Cocaine and other abused stimulants, such as
the amphetamines and methylphenidate, have numerous structural
and neuropharmacological dissimilarities, but all produce, by means
of neurotransmitter reuptake inhibition, activation of dopaminergic
and noradrenergic pathways associated with mood and activation of
serotonergic systems associated with mood and arousal. Manipula-
tions of serotonergic systems modulate the reward potency of both
Downloaded from www.sciencemag.org on October 5, 2012
cocaine and amphetamine in animals, and agents that activate
dopaminergic or noradrenergic systems, or both, but are devoid of
strong serotonergic activity (benztropine, trihexyphenidyl, nomifen-
sine, L-dopa/carbidopa) are neither intensively abused by humans
or, when administered systemically rather than directly into the
CNS, avidly self-administered by animals (29). Pure serotonin
agonists and antagonists lacking acute dopaminergic activity, includ-
ing those with selective activity at specific serotonin receptor
subtypes, also do not produce reward or abuse. Hence, the author
proposes that augmentation of the effects of dopamine reuptake
blockade probably occurs by way of other inputs (serotonergic or
others) to dopaminergic reward pathways to produce sufficient
dopaminergic activation to cause stimulant euphoria. For example,
serotonergic synapses exist on the cell body of dopaminergic
neurons in the midbrain (AlO) that may, in turn, regulate activation
thresholds in reward cells. Cocaine-induced serotonin reuptake
inhibition could magnify cocaine-induced perturbations in the
dopaminergic reward system to induce euphoria. If so, blockade or
perturbation of serotonin or other augmenting systems might block
cocaine effects while minimally influencing normal reward transmis-
sion and, unlike the dopamine receptor-blocking neuroleptics,
averting exacerbation of anhedonia.
Neuroadaptation to Long-Term Cocaine
Abuse
The response of the nervous system to persistent, drug-induced
neurochemical perturbation is a compensatory adaptation in the
perturbed systems. Dysregulation follows adaptation when the drug
is not present. Despite this, until recently it had been assumed that
neuroadaptation does not occur in cocaine abuse. Gawin, El-
linwood, and Kleber have hypothesized that high-dose cocaine use
over long periods of time generates sustained neurophysiological
changes in brain systems that regulate only psychological processes,
particularly hedonic responsivity or pleasure (3, 16). Changes in
these neurophysiological systems produce a true physiological ad-
diction and withdrawal, but one whose clinical expression appears
solely psychological (3).
Intracranial electrical self-stimulation (ICSS) of brain reward sites
in animals provides a model for human pleasure. Chronic cocaine
and amphetamine administration appears to decrease ICSS reward
indices and to increase the threshold voltage required to elicit ICSS
ARTICLES 1583
in dopaminergic reward areas such as the nucleus accumbens (24).
These ICSS decrements imply that brain reward regions affected by
cocaine are subsensitive or down-regulated. They are consistent with
clinical observations of protracted anhedonia in abstinent cocaine
abusers and with behavioral depression after cocaine use in animals.
The neurophysiological dysregulations that would be consistent
with these ICSS data have not been clearly established, but could
take place at several sites. Central dopaminergic, a-adrenergic, and
3-adrenergic receptor supersensitivity has been demonstrated after
long-term cocaine administration in animals (30). Receptor alter-
ations could produce ICSS decrements in animals and anhedonia in
humans (30). It has only recently been possible to differentiate
presynaptic and postsynaptic receptor changes and Dl, D2, D3, and
D4 dopamine receptors. A primary effect ofdecreased dopaminergic
reward transmission would produce a secondary increase in postsyn-
aptic receptor sensitivity. Disproportionate D2 autoreceptor super-
sensitivity (that is, greater functional autoreceptor supersensitivity
than postsynaptic D2 receptor supersensitivity) would decrease
dopaminergic neurotransmission. Such changes have been demon-
strated in animals with continuous but not intermittent long-term
administration of cocaine and amphetamine (31). Alternatively,
other feedback mechanisms could influence firing and activation
thresholds, or other parameters of transmitter release, in dopami-
nergic reward nuclei. These include possible neurotoxic degenera-
tion of dopaminergic reward neurons (3, 32). Dl receptor-mediat-
ed feedback innervation to the cell bodies of reward neurons, and
multiple possible feedback loops involving serotonergic, noradren-
ergic, enkephalinergic, and GABaergic synapses (GABA, -y-ami-
nobutyric acid), among others. Pharmacological interventions at
each of these sites influence cocaine administration in preliminary
preclinical or clinical reports. The predominant pharmacological
treatment thus far investigated in therapy for cocaine addition (the
tricyclic antidepressants) has multiple effects that are opposite to
those of long-term cocaine use: applied on a long-term basis, the
antidepressants increase ICSS reward sensitivity, neuronal firing
rates, and synaptic dopamine concentrations in dopaminergic re-
ward neurons (3, 30). They may also induce dopaminergic autore-
ceptor subsensitivity, and they may have antianhedonic effects in
unipolar depression (3, 30).
The appropriateness of generalization from long-term cocaine
administration studies in animals to human cocaine withdrawal
anhedonia is uncertain. Almost all animal research on long-term
administration of cocaine or other stimulants has used drug admin-
istration paradigms that do not reflect human abuse patterns. For
example, these animal studies usually utilize daily intraperitoneal
administration, producing a slowly increasing single peak in dopa-
minergic facilitation each day, and minimal sleep disruption. This
does not coincide with the multidose binges, perturbed diurnal
cyclicity, and multiple-day cocaine-free intervals characteristic of
human abusers seeking abstinence. Long-term cocaine administra-
tion in animal studies varies in duration from 5 days to 3 months
and may not have consequences similar to multiple years of human
abuse. Further, it is unclear how species differences in life-span
should be interpreted in extrapolation of these studies to humans.
Neurochemical studies in cocaine abusers seeking treatment are
also limited. Several investigations have assessed peripheral neuro-
endocrine and neurotransmitter metabolite indices of central
dopaminergic functioning in treatment-seeking cocaine abusers.
Such studies are potentially confounded by multiple sources: con-
current medical or psychiatric disorders, the intensity and time
course of antecedent cocaine use, other drugs abused, questionable
reliability of details of self-reports, diet, duration of abstinence, and
genetic heterogeneity, among many others. The complexities pre-
sented by differences in antecedent history can, at present, be
1584
resolved only by studies of long-term cocaine administration in
animals in which these variables are systematically regulated.
Despite such limitations, neurochemical studies based on periph-
eral indices of dopaminergic functioning and electroencephalo-
graphic studies in street abusers suggest that neuroadaptation occurs
in street cocaine abuse (18, 20, 33). Both homovanillic acid (HVA),
the principle metabolite of dopamine in humans, and the neurohor-
mone prolactin (inhibited by tuberoinfundibular dopamine release)
have been assessed. Basal prolactin and HVA concentrations in
cocaine abusers are either unchanged or reflect decreased dopami-
nergic functioning (34). However, three studies now indicate that
transiently increased dopaminergic functioning is associated with
resurgence of craving for cocaine (18, 33). This is consistent with
microdialysis studies in animals indicating that conditioned cues and
nondrug stimuli that indicate craving (exposure of male rodents,
without access, to sexually receptive females) are associated with
increased dopamine release, of substantially less magnitude than that
caused by cocaine, in the nucleus accumbens (35). It is possible that
such fluctuations in dopaminergic activity occur broadly enough to
be reflected in less sensitive peripheral assessments in humans.
In vivo imaging of human brain functioning can directly measure
Downloaded from www.sciencemag.org on October 5, 2012
metabolism and neurophysiology in cocaine abusers (18). The
spatial and temporal resolution of PET scanning, single photon
emission computerized tomography, magnetic resonance spectros-
copy, and methods of assessing regional cerebral blood flow appear
to be insufficient for precise demonstration of alterations in reward
function in human abusers. For example, diminished reward func-
tion could occur only in the nucleus accumbens, which is beyond the
spatial resolution of current methods; and the period of craving that
follows a conditioned cue may be associated with very brief activa-
tion or deactivation of reward pathways that may not be detectable
because of the relatively lengthy acquisition time of these imaging
techniques. Simple practical problems also exist. Because of the
unpredictable time course of cocaine abstinence symptoms, imaging
cannot be scheduled to coincide with specific clinical states, and
keeping equipment and personnel at ready is implausible.
Recovery and Treatment
Data from treatment programs using different therapeutic ap-
proaches indicate that outpatient cocaine abuse treatment can be
successful. From 30 to 90% of abusers remaining in outpatient
treatment programs cease cocaine use (36). Samples vary widely in
degree of abuse severity, psychosocial resources, route of adminis-
tration, and magnitude and pattern of cocaine and other drug
exposure, so they are not readily comparable. Further, rigorous
design is usually lacking, so little evidence exists to suggest that any
one treatment approach is superior (3). Also, despite superficial
differences, strong commonalities characterize all current treatment
strategies. Initial efforts focus on retaining the abuser in treatment
and disrupting the cycles of binges; subsequent efforts focus on
relapse prevention (3).
There is controversy regarding the relative therapeutic value of,
and indications for, most components of cocaine abuse treatment.
These include, but are not limited to, hospital versus outpatient
cocaine abuse treatment; types of treatment techniques; use, selec-
tion, dosage, or duration of adjunctive medication; lengths of
isolation from and rapidity of latter immersion in the environment
in which cocaine is available; and most other variables in treatment
design. These areas require assessment in controlled, random-
assignment treatment studies.
Recent double-blind comparisons of medication treatment and
abstinence initiation in outpatients have demonstrated that hetero-
SCIENCE, VOL. 251
 4.0 Fig. 2. Desipramine fa- ponents. Initial clinical investigations of how the "normal euphoria
cilitation of initial co- of a healthy person" (7) becomes transformed into the source of
2 caine abstinence. End-
(D 3.0 __
t
point analysis of
"mad craving" (6) have resulted in the promise and initial actualiza-
reported cocaine use tion of advances in pharmacological and psychotherapeutic treat-
(grams) during treat- ment of cocaine addiction. These advances have been founded in
=2.0 ment of outpatient co- basic animal research, thus underscoring the value of basic research
directed at unraveling the neurophysiological mysteries of human
caine abusers with desi-
pramine hydrochloride experiences of pleasure and pain.
o 1.0
0
(2.5 mg per kilogram of
body weight per day, n
= 24, solid line), lithium
REFERENCES AND NOTES
0 0 1 2 3 4 5 6 7 carbonate (17.5 mg/kg,
ime after treatment onset (week) or 1. World Health Organization (WHO) Expert Committee on Addiction-Producing
nplacebo = 24,line),
= 24, (ndashed dotted Drugs, Bull. WHO 39, 225 (1981).
line). [Reprinted from Gawin et al. (37) with permission, Arch. Gen. 2. B. J. Rounsaville, R. Spitzer, J. B. Williams, Am.J. Psychiatry 143, 463 (1986).
Psychiatry] 3. F. Gawin and E. H. Ellinwood, N. Engi. J. Med. 318, 1173 (1988); Treatment of
Psychiatric Disorders: A Task Force Report of the American Psychiatric Association
(American Psychiatric Association, Washington, DC, 1989).
cyclic antidepressants (desipramine and imipramine), which do not 4. F. H. Gawin, D. Allen, B. Humblestone, Arch. Gen. Psychiatry 46, 322 (1989).
produce dependence or addictive use patterns, diminish cocaine use 5. E. H. Adams and N. J. Kozel, in NatI. Inst. Drug Abuse (NIDA) Res. Monogr. 61,
(1985), pp. 35-49.
and craving and improve initial treatment outcome in the first 6. The psychic effect of cocaine consists of exhilaration and lasting euphoria ...
months of treatment (Figs. 2 and 3) (4, 37). Multiple uncontrolled which does not differ from the normal euphoria ofa healthy person.... Absolutely
no craving for cocaine appears after the first, or repeated, taking of the drug" [S.
reports also exist ofdecreases in craving induced by other agents that Freud, Centralbl. Gesammt Ther. 2, 289(1884), pp. 300-302].

Downloaded from www.sciencemag.org on October 5, 2012

are hypothesized to normalize reward function, and multiple dou-
ble-blind trials of similar pharmacological treatment for cocaine
craving are in progress. Efforts to block the effects of cocaine appear
less promising. Dopamine blocking agents may block cocaine
effects, but apparently do so only at doses that so markedly amplify
anhedonia that such treatment is counter-therapeutic (28, 38).
Reports of other agents that may decrease self-administration of
cocaine in animals have been difficult to interpret. Buprenorphine,
for example, was reported to diminish cocaine administration in
primates by presumed decreases in reward potency (39). However,
increases in reward also diminish frequency of administration in
animals, and rodent studies are consistent with buprenorphine
augmentation, rather than blockage, of the effects of cocaine (40).
The feasibility of systematic, controlled psychotherapy research
for cocaine abuse is demonstrated by promising assessments of
systematic desensitization of cues eliciting conditioned cocaine
craving, but only preliminary clinical trials have appeared (43).
In-depth studies of the outcome of systematic desensitization are in
progress, as are studies contrasting psychotherapies with and with-
out pharmacotherapies that are similar to prior studies of severe
depression treatment in non-drug users.
Conclusions
Cocaine addiction has evolved within one decade from a virtually
nonexistent problem to a complexly regulated disorder with inter-
woven environmental, psychological, and neurophysiological com-
8 Fig. 3. Effect of chronic
desipramine treatment
on cocaine-induced co-
caine craving. Intrave-
am nous cocaine (0.5 mg
per kilogram of body
0 weight) or cocaine place-
bo infused in five sub-
c0) jects after long-term
0, treatment with desipra-
mine hydrochloride
0 (150 mg/day for greater
than 10 days) or placebo
maintenance. (C, place-
bo/placebo; *, placebo/
0 100 200 desipramine; U, cocaine/
Time after infusion (min) placebo; and 0, cocaine/
desipramine)
29 MARCH 1991
7. By Lewin in 1887: "They give all that they possess, in order to indulge their mad
craving" [L. Lewin, Phantastica (Stilke, Berlin, 1924), p. 80]; H. W. Meier, Der
Kokainismus (Verlag, Leipzig, Germany, 1926).
8. E. H. Ellinwood, in Drug Use: Epidemiological and Sociological Approaches, F.
Josephson and E. Carroll, Eds. (Hemisphere, Washington, DC, 1974), pp.
303-329.
9. 'Chronic cocaine abuse does not usually appear as a medical problem" [L.
Grinspoon, and J. B. Bakalar, in Comprehensive Textbook of Psychiatry, H. I.
Kaplan, A. M. Freedman, B. J. Sadock, Eds. (Williams & Wilkins, Baltimore,
1980)]; National Commission on Marijuana and Drug Abuse, Drug Use in
America: Problems in Perspective (Government Printing Office, Washington, DC,
1973).
10. "Cocaine is, from a public health perspective ... the drug of greatest national
health concern" [J. Grabowski, NIDA Res. Monogr. 50 (1984), p. viii]; J. F. Jekel
et al., Lancet i, 459 (1986); F. H. Gawin and H. D. Kleber, Arch. Gen. Psychiatry
43, 107 (1986); R. K. Siegel,]. Psychoact. Drugs 14, 321 (1982).
11. F. H. Gawin and H. D. Kleber, NIDA Res. Monogr. 61 (1985), p. 182.
12. C. Johanson and M. Fischman, Pharmacol. Rev. 41, 3 (1989).
13. C. Van Dyke et al., Int. J. Psychiatry Med. 12, 1 (1982); E. H. Ellinwood, in
Psychopharmacology in the Practice ofMedicine, M. E. Jarvik, Ed. (Appleton-Century-
Crofts, New York, 1977), pp. 467-479.
14. M. Scherer, K. Kumor, E. Cone, J. Jaffe, Arch. Gen. Psychiatry, 45, 673 (1988).
15. E. H. Ellinwood and W. M. Petrie, in DrugAbuse: Clinical and Basic Aspects, S. N.
Pradham, Ed. (Mosby, New York, 1977), pp. 248-262.
16. F. H. Gawin and H. D. Kleber, Arch. Gen. Psychiatry, 43, 107 (1986).
17. P. H. Connell, in Drugs and Youth, J. R. Wittenborn et al., Eds. (Thomas,
Springfield, IL, 1970).
18. H. Kalsa, D. Anglin, and F. Gawin (manuscript in preparation) describe consistent
findings in the first large-scale investigation of withdrawal symptoms in 240
outpatients. In smaller hospitalized populations, S. Martin, V. Yeragani, R. Lohdi,
and M. P. Galloway [Biol. Psychiatry 26, 356 (1989)]; S. Hollanderet al. 33, 1611
[Psychiatry. Res. (1990)]; N. D. Volkow et al. [Am. J. Psychiatry 147, 719
(1990)]; and N. D. Volkow, N. Mullani, K. Gould, S. Adler, and K. Krajewski
[Br. J. Psychiatry 152, 641 (1988)] have reported findings consistent with prior
outpatient evaluations, but W. Weddington et al. [Arch. Gen. Psychiatry, 47, 861
(1990)] have not. W. Wolverton and M. S. Kleven [Psychopharmacology 94, 288
(1988)]; D. M. Woods and H. Lal [Life Sci. 41,1431 (1987)]; and M. E. Carrol
and S. T. Lac [ibid. 40, 2183 (1987)] have also described consistent findings on
cocaine discontinuation in animals.
19. S. A. Satel and F. H. Gawin, Am. J. Psychiatry, in press.
20. R. Watson, E. Hartmann, J. J. Schildkraut, ibid. 129, 263 (1972).
21. Only the Fawcett anhedonia scale has been validated as an index for assessing
anhedonia in nonpsychotic populations, but in alcohol abusers (J. Fawcett,
personal communication) and cocaine abusers (F. H. Gawin, unpublished obser-
vations), construct validity for this instrument appears to be absent.
22. A. J. Giannini and C. Sangdahl, Int. J. Psychiatry Med. 15, 415 (1985); C. P.
O'Brien et al.,J. Clin. Psychiatry 49 (suppl.), 17 (1988).
23. H. Utena, Prog. Brain Res. 21, 1902 (1962).
24. Reviewed in F. Gawin and E. H. Ellinwood, New Engl.J. Med. 318, 1173 (1988).
See also A. Markou, C. Hubner, and G. Koob [Soc. Neurosci. Abstr. 14, 660
(1988)], and L. Kokkinidis and B. D. McCarter [Pharmacol. Biochem. Behav. 36,
463 (1990)].
25. R. Wise,J. Abnormal Psychol. 97, 118 (1988); G. F. Koob and F. E. Bloom,
Science 242, 715 (1988).
26. M. C. Ritz, R. J. Lamb, S. R. Goldberg, M. J. Kuhar, Science 237, 1219 (1987);
M. Reith, H. Sershen, D. L. Allen, A. Lajtha, Mol. Pharmacol. 23, 600 (1983); R.
Jones, NIDA Res. Monogr. 50 (1984), p. 182.
27. M. Memo, S. Pradham, I. Hanbaur, Neuropharmacology 20, 1145 (1981).
28. C. Spyraki, H. C. Fibiger, A. C. Phillips, Brain Res. 253, 195 (1982); F. H. Gawin,
Psychopharmacology 90, 142 (1986).
ARTICLES 1585
29. M. E. Carrol, S. Lac, M. Asencio, R. Kragh, Psychopharmacology 100, 293 (1990);
Pharmacol. Biochem. Behav. 35, 237 (1990); J. G. Modell, R. Tandon, T. P.
Beresford,J. Clin. Psychopharmacol. 9, 347 (1989).
30. F. Gawin and H. Kleber, Arch. Gen. Psychiatry 42, 903 (1984).
31. L. P. Dwoskin, J. Perns, R. Yasuda, K. Philpot, and N. Zahniser [Life Sci. 42, 255
(1988)]; and D. J. Henry, M. Greene, and F. White U. Pharmacol. Exp. Ther. 251,
901 (1989)] document autoreceptor subsensitivity with intermittent administra-
tion of cocaine, but see M. Reith, M. Benuck, and A. Lathja [ibid. 243, 281
(1987)]; and Kokkinidis and McCarter [in (24)] for a demonstration of the
importance of the dosing pattern in cocaine neuroadaptation. Increased effects of
single-dose cocaine administration after repeated cocaine administrations, or
sensitization, can also occur. The relevance of sensitization for human cocaine
addiction and craving is questionable. The animal literature contains multiple
reports describing the occurrence of sensitization with regularly repeated (for
example, daily) small cocaine doses. The only similar clinical phenomenon is
cocaine-induced paranoia, which occurs only in a subpopulation of users (S. A.
Satel and F. H. Gawin, Am. J. Psychiatry, in press) and which does not directly
contribute to cocaine addiction.
32. R. J. Wyatt, R. Fawcett, F. Karoum, paper presented at the 26th Annual Meeting
of the American College of Neuropsychopharmacology, San Juan, Puerto Rico,
7-11 December 1987; M. Trulson, S. Babb, J. C. Joe, J. D. Raese, Exp. Neurol.
94, 744 (1986).
33. F. H. Gawin and H. D. Kleber, Br.J. Psychiatry 147, 569 (1985).
34. C. Dackis and M. Gold, Neurosci. Behav. Res. 9, 469 (1985); I. Extein et al., paper
presented at the 140th Meeting of the American Psychiatric Association, Chicago,
" Whatever it was, it was around during the ice-age. "
1586
IL, 10-15 May 1987; J. Mendelson et al.,Am.J. Psychiatry 145, 553 (1988); N.
S. Pilotte, L. G. Sharpe, E. M. Dax, Brain Res. 512, 107 (1990); C. M. Schwarz,
K. Breen, F. Leone, Am.J. Psychiatry 147, 777 (1990).
35. P. W. Kalivas and P. Duffy, Synapse 5, 48 (1990); J. R. Blackburn, A. G. Phillips,
A. Jakuboic, H. C. Fibiger, Behav. Neurosci. 103, 15 (1989).
36. F. H. Gawin and H. D. Kleber, in Cocaine: Clinical and Biobehavioral Aspects, S.
Fisher, B. Raskin, R. Uhlenhuth, Eds. (Oxford Univ. Press, New York, 1986), pp.
169-87; C. P. O'Brien et al., NIDA Res. Monogr. 95 (1989), p. 78.
37. A. Gianninni, D. Malone, M. Gianninni, W. Price, R. Loiselle,. Clin. Pharmacol.
26, 211 (1986); N. Nunes, F. Quitkin, D. Klein, in preparation; F. H. Gawin et
al., Arch. Gen. Psychiatry 46, 117 (1989); S. L. McElroy et al., NIDA Res.
Monogr. 95 (1989), p. 57.
38. M. Sherer, K. Kumor, J. Jaffe, Psychiatry Res. 27, 117 (1989).
39. N. K. Mello, J. H. Mendelson, M. P. Bree, S. E. Lukas, Science 245, 859
(1989).
40. E. E. Brown et al., Soc. Neurosci. Abstr. 16, 251 (1990).
41. A. Childress, R. Ehrman, A. McLellan, C. O'Brien, NIDA Res. Monogr. 81 (1988),
p. 74.
44. Supported by NIDA grants P50-04060, R18-DA06082, and R01-DA06289, by
the Friends of Medical Research and by the Foundation for the International
Exchange of Scholars. I thank D. Allen, R. Byck, E. Ellinwood, H. Fibiger, B.
Humblestone, J. Jaffe, J. Jekel, H. Kleber, G. Koob, P. Jatlow, C. O'Brien, N.
Nunes, R. Schuster, R. Weiss, and R. Wise for stimulating and illuminating
discussions. The opinions expressed are, however, solely the responsibility of the
author.
Downloaded from www.sciencemag.org on October 5, 2012
SCIENCE, VOL. 251